Scientists Challenge Nikon Microscopy Competition Winner Over AI-Enhanced Video With Biological Inaccuracies

Scientists Question Biological Accuracy in Award-Winning Microscopy Video

Ning Xu, an optical engineer at the National University of Singapore, won Nikon's 16th annual Small World in Motion competition on September 15 with a video depicting cilia in lung tissue from a child with primary ciliary dyskinesia1

1

. Within days, the entry ignited an AI controversy as microscopy experts raised concerns about biologically implausible features in the scientific visualization2

2

. Edward Phelps, a bioengineering researcher at the University of Florida in Gainesville, identified purple structures resembling mitochondria that don't occur naturally in biology at the size shown, while blue structures appeared to resemble nuclei but didn't behave like them1

1

. The red structures remained entirely unidentified.

Nikon Re-Reviewing Winner After SynthID Watermark Discovery

The Small World in Motion competition explicitly prohibits AI-generated videos, yet Ian Donovan, an MD/PhD student at UT Southwestern Medical Center, discovered a Google SynthID watermark when running the video through Google Gemini3

3

. This watermark, pioneered by Google's DeepMind research lab, indicates AI-generated content. Following the discovery, Nikon Instruments updated its blog post on September 22 to acknowledge that Xu used an unsupervised AI model for post-processing1

1

. Nikon is now carefully re-reviewing the submission, with Xu providing detailed technical documentation outlining microscopy equipment, imaging methods, and processing techniques2

2

.

Defender Claims AI Enhanced Visual Presentation, Not Core Data

Xu responded to concerns on LinkedIn, stating his team applied post-processing to enhance visual presentation of regions below the cilia "without making anatomical claims about what those rendered features represent"1

1

. He emphasized that AI was not used to generate the experimental movie, the cilia, or their motion, but was "applied afterwards to the reconstructed grayscale data to distinguish and color structures with similar morphology"2

2

. Xu explained the video was prepared for a competition celebrating microscopy's beauty, so the team wanted the final presentation to be visually engaging as well as scientifically interesting1

1

. However, Xu's LinkedIn profile has since been deleted3

3

.

Scientific Community Demands Trust in Experimental Data

Melanie White, a developmental biologist at the University of Queensland in Brisbane, Australia, stressed that scientific images are data, not just illustrations, and researchers need to trust that what they're seeing is grounded in underlying measurement1

1

. Markus Sauer, who studies super-resolution microscopy at the University of Würzburg in Germany, clarified that using AI tools for scientific visualization isn't automatically problematic, but becomes an issue when AI-enhanced videos misrepresent, exaggerate, or alter experimental findings1

1

. Former Small World in Motion judge Andrew Moore compared the situation to old photographs restored with AI, noting that while high-resolution colorized photos may look nice, if your grandmother was face-swapped, it would be unsettling2

2

.

AI Confabulations Raise Ethical Questions About Post-Processing

The ethical use of AI in scientific images has become central to this controversy. White noted it's unclear how faithfully the video represented information captured by the microscope, with the possibility that the AI tool introduced biological structures that don't actually exist in the data1

1

. Fifth-place winner Patrick Hickey told the BBC that microscopists don't get many chances to showcase their work, which often has abstract and artistic qualities2

2

. Scientific guidelines from 2010 allow for stained samples or colorized images, provided changes are disclosed and described, but these guidelines predate generative AI's ability to fabricate pixels outright2

2

.

Resolution Hinges on Raw Grayscale Video Release

Scientists have repeatedly stated the controversy could be resolved if Xu shares his raw grayscale video, which would allow independent verification of the experimental data before AI post-processing3

3

. Sarah Goetz captured the scientific community's sentiment on Bluesky, writing that if AI was used to generate synthetic images, "there is something profoundly cynical and sad about using synthetic images to cheat in a contest that is supposed to be about how beautiful and cool biology is"3

3

. The case highlights growing concerns about maintaining scientific integrity when AI tools blur the line between enhancement and fabrication in cilia in lung tissue imaging and other biological research.